https://ir.cut.ac.zw:8080/xmlui/handle/123456789/724 Research articles 2026-06-06T01:33:31Z https://ir.cut.ac.zw:8080/xmlui/handle/123456789/739 Modelling and Forecasting the Value of Special Drawing Rights: An ARIMA Approach Samambgwa, Henry; Musora, Thomas; Mamutse, Dennis In 1969, the International Monetary Fund (IMF) introduced Special Drawing Rights (SDR) as a financial instrument to supplement currency reserves of member states. SDR allow members to draw money in a currency of their choosing. 2025 IMF data valued SDR held by member states at over USD 660 billion. SDR are, thus, a critical financial indicator of significant potential impact on global economic stability. This study analysed SDR prices from November 2015 to October 2025. The Minimal Information Criterion and the Bayesian Information Criterion were applied to compare ARIMA models, and the ARIMA (1,1,0) emerged as the best fit. Model diagnostics confirmed that no validity assumptions were violated. Observed values were regressed against fitted values. The R^2 value was over 90%, indicating a very strong linear relationship, which is very plausible. The model was then used to forecast SDR prices for the months from November 2025 to April 2026. The findings revealed a slight decline in SDR prices in the forecasted period. These insights have a significant impact on IMF member states, investors and international economic policy makers. 2025-12-12T00:00:00Z https://ir.cut.ac.zw:8080/xmlui/handle/123456789/738 Deriving mathematical models from neural networks: A method for deducing individual effects of factors on a response variable Samambgwa, Henry; Musora, Thomas; Kamusha, Joseph This research outlines a novel approach to obtaining mathematical models from neural networks. The target scenario is one where a response variable depends on a number of factors, each factor has an effect which is a function of the factor and the response variable is the sum of the effects of the factors. A neural network was trained such that response values were generated from factor values. It was assumed that each effect was zero when the underlying factor was set to zero. The effect of a factor could be isolated by setting all other factors to zero, so that the response value became equal to the effect of the factor being isolated. In that way each effect was isolated and then modelled as a function of the factor. Thus, the technique was developed, for modelling a response variable as a function of its input factors. 2025-12-01T00:00:00Z https://ir.cut.ac.zw:8080/xmlui/handle/123456789/737 Automating Upper Triangular Matrix Neutralisation for Minimal Error Samambgwa, Henry; Musora, Thomas This study develops and tests a MATLAB program for neutralising systems with upper triangular coefficient matrices. Augmented matrices for systems of linear equations are reduced to produce a neutral (identity matrix) left hand side, at which point the right hand side has the solutions for the system. A strategy for minimal error is pursued, where elimination operations must be completed in one step for each number. The strategy was expressed in pseudocode, a flowchart developed, and then the program was coded in MATLAB. The program was tested for four sample augmented matrices of varying sizes. The results demonstrated that the program was 100% accurate. The technique can be applied to any square matrix system, since any square matrix can be expressed as an upper triangular matrix. The automation of matrix operations and achievement of 100% accuracy allows for the use of the program for very large data sets, extending the potential for research with accurate finding. 2025-11-18T00:00:00Z https://ir.cut.ac.zw:8080/xmlui/handle/123456789/735 A systematic approach to extracting multivariate polynomial models from neural networks Samambgwa, Henry; Musora, Thomas This study develops a systematic approach to extract multivariate polynomial models from neural networks. A neural network is trained using a random dataset generated using a bivariate polynomial. Numerical values obtained from simulating the neural network are used to estimate the partial derivatives with respect to each input variable with increasing order of partial differentiation until the derivatives become zero. In that way the highest power required in the model is identified for each input variable. A general form of the multivariate polynomial model is written with all possible combinations of powers of the input variables. An appropriate set of corresponding partial differential operators is identified so as to produce a system of partial differential equations from the general polynomial model. The corresponding partial derivatives are estimated numerically and substituted into the equations. On solving the equations, it was found that the method correctly estimates the appropriate parameters for the multivariate polynomial model. 2025-11-12T00:00:00Z